Computational SpikerBox (CSB)

The Computational SpikerBox (CSB) is a hardware-based lab platform that introduces students to computational neuroscience. Developed by the engineers at Backyard Brains, the CSB enables students to explore how ion channels of neurons function using mathematical models. Students interact with physical knobs” to open, shut, or operate Na++, K++, and other Ion Channels. Optional Addon boards allow students to engage in virtual electrophysiology, determining which neuron type the electrode picks up by interacting with various sensors (touch, vibration, light, heat, etc) — bridging neuroscience theory with hands-on experience.

Overview

• What It Does

  1. Hodgkin-Huxley-Serbe Model: A spiking neuron model where students can twist knobs (ion channel conductances) to see how poisons and venums can affect the nervous system.
  2. Izhikevich + TinyML Model: An “EPhys Neuron Identification Game” that integrates machine learning to classify sensor data and drive a simplified neuron model in real-time.

• Who It’s For

  • Teachers can run a single CSBox in front of the class, or
  • Each student can have their own CSBox (if resources allow) to dive deeper into computational models, sensor readings, and even machine learning.

• Educational Goals

  • Demonstrate how theoretical neuron equations (Hodgkin-Huxley-Serbe, Izhikevich) can be interactively explored.
  • Provide a hands-on introduction to machine learning classification, bridging neuroscience, electronics, and STEM.

---

Repository Contents

• CSBv1.1.ino
  This is the latest, feature-complete release. It implements the two main modes (Hodgkin-Huxley-Serbe for biophysical spiking, and Izhikevich + TinyML for the “Neuron Identification Game”) as described above.

• ei-empanada-arduino-1.0.X.zip
  This is the latest Machine Learning header. Needed for Izhikevich model and Neuron guessing game

• MLReadin
  A specialized sketch for reading sensor data into Edge Impulse (TinyML) for model training.

• ETBiophysical
  An experimental or future direction, converting a Python Colab-style biophysical model into Arduino code.

---

Team & Acknowledgments

Team: Etienne, Chethan, Hatch, Max, with Greg
Advice from: Stan

The Izhikevich model portion is adapted from the original paper (2003 IEEE) and informed by Spikeling v1.1 by T. Baden (Sussex Neuroscience).

---

Helpful Links

1. CSBox High School Lesson Plan
   Google Doc

2. Empanada Machine Learning Model
   Edge Impulse Public Project

3. Edge Impulse Intro from Grove Article
   Seeed Wiki Page

4. Edge Impulse Intro from Grove Video
   YouTube Link

5. Arduino Library Installation After Model Creation
   Edge Impulse Docs

---

Usage & Getting Started

1. Hardware Setup

   • Connect your CSBox main board with potentiometers (for Hodgkin-Huxley-Serbe) and any add-on boards (like “SKIN”) for the ML-based neuron ID game.
   • Make sure you have a compatible microcontroller (e.g., Arduino-type board) and all required sensors (FSR, Temp, Light).

2. Software & Libraries

   • Install the Arduino IDE (or PlatformIO).
   • Ensure you have the Adafruit NeoPixel library and any relevant libraries from Edge Impulse installed.

3. Flashing CSBv1.ino

   • Download both CSBv1.ino and ei-empanada-arduino-1.0.X.zip
   • Open CSBv1.ino in the Arduino IDE.
   • Select your board and COM port.
   • Go to Sketch -> Include Library -> Add .ZIP library
   • Add ei-empanada-arduino-1.0.X.zip
   • Compile and upload the sketch.

4. Running & Observing

   • Mode 1 (Hodgkin–Huxley): If no add-on board is detected (attachment sense pin < threshold), you’ll see the spiking behavior controlled by three knobs.
   • Mode 2 (Izhikevich + TinyML): If the SKIN add-on board is attached (attachment sense pin >= threshold), the sketch runs a classification on sensor data to decide whether to “fire” the neuron model.
   • Open Spike Recorder to see the model play out in real time -(Optional debug mode ) Monitor the Serial output to see debug info or classification confidence.

5. Lesson Integration

   • Check the High School Lesson Plan for ideas on how to structure classroom labs around these models and sensors.
   • Encourage students to experiment with sensor inputs, observe changes in spiking behaviors, and track their hypotheses versus actual model outcomes.

---

Future Directions

• Additional Models: You can import new or extended neuron models (e.g., Hodgkin–Huxley, heart pacemaker cells).
• Deeper ML: Train your own classifiers with new sensor data (using Edge Impulse).
• Advanced Lab Activities: Introduce noise, bursting phenomena, or combined synaptic inputs for more complex lessons.

---

License:
CC BY-NC 4.0 – This project is free for educational and non-commercial use.

Happy Spiking!